It has long been though that injuries during the first states of brain development either kill the embryo, result in neural tube defects, or are followed by complete recovery of the CNS, making this period of little interest to neurotoxicologists. However, new evidence suggests that injury during the earliest stages of brain development may be of great importance. Studies of mice transgenic for targeted disruptions of genes involved in segmentation of the hindbrain reveal that whole rhombomeres can be lost without obvious changes in the development of the rest of the CNS. Thus, it is not true that massive lesions of the neural tube either stop development or recover. Are there human conditions of unknown cause with brain stem lesions due to injury during neural tube closure? We propose that this pattern is the one that leads to autism. In a recent report, thalidomide cases exposed during neural tube closure exhibited an autism rate of over 30%, along wit neurological symptoms of brain stem injury. Using valproic acid, we have reproduced features of the thalidomide cases in a rat model. These animal also have abnormalities of the cerebellum similar to those reported in human cases of autism. To test the hypothesis that autism arises from injury to the closing neural tube, we examined an autistic autopsy case and found evidence of an early injury to the brain stem. Because the neuroanatomy of the autopsy case reproduces several features of the Hoxa-1 knockout mouse, we now propose to test the hypothesis that this gene or related early developmental genes are the source of the genetic liability in autism. We shall sequence the Hoxa-1 gene in autopsy cases and in familial and sporadic living cases of autism, looking for a mutation. Cases of autism will be characterized as t symptoms of the disorder, cognition, language, neurology (including motor function), and physical anomalies. When a mutation is identified, we shall proceed to study the pattern of inheritance of the gene and of the symptoms, testing family members of cases with mutations genetically and behaviorally. Unlike other attempts to identify a genetic cause of autism, this one is based on data about the developmental origin and neuroanatomical phenotype responsible for some cases of the disease. Identification of a mutation underlying autism would allow a clinical test r the deficit and the creation of transgenic animals in which teratogens coul be tested for the potential to trigger the disease.